In verbal communication, a receiver often needs to resolve referential ambiguity. This study set two experimental conditions to separate the possibility of local correspondence based on the persisting strategy of reference assignment from that of more flexible reference skills. A total of 139 three-year-old and five-year-old children engaged in referencing names and colours in response to ambiguous implicit questions. The double-referent condition involved a reference assignment task, which alternated explicit questions and implicit questions. In the single-referent condition, implicit questions appeared at different positions within a series of interactions. The results demonstrated no difference between the conditions, indicating that the difficulty of disambiguation was not due to local correspondences such as "this"-referent mappings in a particular context. This suggests that the study's procedure appropriately probed inferential competence to enable retrospective disambiguation, thus providing a potential tool for assessing various aspects of pragmatic interpretation.